pyODK¶

An API client for the ODK Central API. Use it to interact with your data and automate common tasks from Python.

This library aims to make common data analysis and workflow automation tasks as simple as possible by providing clear method names, types, and examples. It also provides convenient access to the full API using HTTP verb methods.

Install¶

The currently supported Python version for pyodk is 3.12. If this is different from the version you use for other projects, consider using pyenv to manage multiple versions of Python.

The currently supported Central version is v2024.1.0. Newer or older Central versions will likely work too, but convenience (non-HTTP) methods assume this version. If you see a 404 error or another server error, please verify the version of your Central server.

From pip¶

pip install pyodk

From source¶

# Get a copy of the repository.
mkdir -P ~/repos/pyodk
cd ~/repos/pyodk
git clone https://github.com/getodk/pyodk.git repo

# Create and activate a virtual environment for the install.
python -m venv venv
source venv/bin/activate

# Install pyodk and its production dependencies.
cd ~/repos/pyodk/repo
pip install -e .

# Leave the virtualenv.
deactivate

Configure¶

The configuration file uses the TOML format. The default file name is .pyodk_config.toml, and the default location is the user home directory. The file name and location can be customised by setting the environment variable PYODK_CONFIG_FILE to some other file path, or by passing the path at init with Client(config_path="my_config.toml"). The expected file structure is as follows:

[central]
base_url = "https://www.example.com"
username = "my_user"
password = "my_password"
default_project_id = 123

Custom configuration file paths¶

The Client is specific to a configuration and cache file. These approximately correspond to the session which the Client represents; it also encourages segregating credentials. These paths can be set by:

Setting environment variables PYODK_CONFIG_FILE and PYODK_CACHE_FILE
Init arguments: Client(config_path="my_config.toml", cache_path="my_cache.toml").

Default project¶

The Client is not specific to a project, but a default project_id can be set by:

A default_project_id in the configuration file.
An init argument: Client(project_id=1).
A property on the client: client.project_id = 1.

Default Identifiers

For each endpoint, a default can be set for key identifiers, so these identifiers are optional in most methods. When the identifier is required, validation ensures that either a default value is set, or a value is specified. E.g.

client.projects.default_project_id = 1
client.forms.default_form_id = "my_form"
client.submissions.default_form_id = "my_form"
client.entities.default_entity_list_name = "my_list"
client.entities.default_project_id = 1

Session cache file¶

The session cache file uses the TOML format. The default file name is .pyodk_cache.toml, and the default location is the user home directory. The file name and location can be customised by setting the environment variable PYODK_CACHE_FILE to some other file path, or by passing the path at init with Client(config_path="my_cache.toml"). This file should not be pre-created as it is used to store a session token after login.

Use¶

To get started with pyODK, build a Client:

from pyodk.client import Client

client = Client()

Authentication is triggered by the first API call on the Client, or by explicitly using Client.open().

Use Client.close() to clean up a client session. Clean up is recommended for long-running scripts, e.g. web apps, etc.

You can also use the Client as a context manager to manage authentication and clean up:

with Client() as client:
    print(client.projects.list())

Learn more in the documentation.

Examples¶

👉 See detailed tutorials in the documentation.

from pyodk.client import Client

client = Client()
projects = client.projects.list()
forms = client.forms.list()
submissions = client.submissions.list(form_id=next(forms).xmlFormId)
form_data = client.submissions.get_table(form_id="birds", project_id=8)
comments = client.submissions.list_comments(form_id=next(forms).xmlFormId, instance_id="uuid:...")
client.forms.update(
  form_id="my_xlsform",
  definition="my_xlsform.xlsx",
  attachments=["fruits.csv", "vegetables.png"],
)
client.close()

Session customization¶

If Session behaviour needs to be customised, for example to set alternative timeouts or retry strategies, etc., then subclass the pyodk.session.Session and provide an instance to the Client constructor, e.g. Client(session=my_session).

Logging¶

Errors raised by pyODK and other messages are logged using the logging standard library. The logger is in the pyodk namespace / hierarchy (e.g pyodk.config, pyodk.endpoints.auth, etc.). The logs can be manipulated from your script / app as follows.

import logging

# Initialise an example basic logging config (writes to stdout/stderr).
logging.basicConfig()
logging.getLogger().setLevel(logging.DEBUG)

# Get a reference to the pyodk logger.
pyodk_log = logging.getLogger("pyodk")

# Receive everything DEBUG level and higher.
pyodk_log.setLevel(logging.DEBUG)
pyodk_log.propagate = True

# Ignore everything below FATAL level.
pyodk_log.setLevel(logging.FATAL)
pyodk_log.propagate = False

Errors raised by pyODK¶

Error types raised by pyODK are found in errors.py, which currently is only the PyODKError. In general this error is raised when:

The pyODK configuration is invalid (missing file, missing fields, etc).
The client method arguments are invalid (missing, wrong type, etc.).
The response from ODK Central indicated and error (HTTP >=400, <600).
The data returned from ODK Central does not have the expected fields or types.

Note that pyODK does not attempt to wrap every possible error condition, so if needed, broader exception handling should be included in your script / app.

Design considerations¶

Our goal with pyODK is to support the most common Central API functionality in an easy-to-use, high-level way. Because we expose HTTP verb methods, we don't feel the need to add explicit support for the whole Central API.

Here are some points to think about when considering adding new methods to pyODK:

Is it common enough to warrant a designed method or can we show examples using the HTTP methods if needed?
For example, we currently consider manipulating form drafts directly out of scope but client.forms.update implicitly creates and publishes a draft
Do people take this action independently or is it always part of reaching some broader goal?
What pyODK class does it best fit in?
For example, form versions are subresources on Central backend but in this library, methods that deal with form versions can be in the forms class directly since we’re not going to expose many of them
How do people talk about the action that’s being performed? How do ODK docs and Central frontend talk about it?
For example, documentation has the concept of "submission edits" so use client.submissions.edit rather than update
Value expressiveness and consistency with ODK concepts over pyODK internal consistency
What actually needs to be commonly configured? Starting by exposing a subset of parameters and naming/typing them carefully is ideal.

Contribute¶

See issues for additions to pyodk that are under consideration. Please file new issues for any functionality you are missing.

Develop¶

Install the source files as described above, then:

pip install -e .[dev]

You can run tests with:

python -m unittest

Testing¶

When adding or updating pyODK functionality, at a minimum add or update corresponding unit tests. The unit tests are filed in tests/endpoints or tests. These tests focus on pyODK functionality, such as ensuring that data de/serialisation works as expected, and that method logic results in the expected call patterns. The unit tests use mocks and static data, which are stored in tests/resources. These data are obtained by making an API call and saving the Python dict returned by response.json() as text.

For interactive testing, debugging, or sanity checking workflows, end-to-end tests are stored in tests/test_client.py. These tests are not run by default because they require access to a live Central server. The ODK team use the Central staging instance https://staging.getodk.cloud/ which is already configured for testing. Below are the steps to set up a new project in Central to be able to run these tests.

Create a test project in Central.
Create a test user in Central. It can be a site-wide Administrator. If it is not an Administrator, assign the user to the project with "Project Manager" privileges, so that forms and submissions in the test project can be uploaded and modified.
Save the user's credentials and the project ID in a .pyodk_config.toml (or equivalent) as described in the above section titled "Configure".
When the tests in test_client.py are run, the test setup method should automatically create a few fixtures for testing with. At a minimum these allow the tests to pass, but can also be used to interactively test or debug.

Release¶

Run all linting and tests.
Draft a new GitHub release with the list of merged PRs.
Check out a release branch from latest upstream master.
Update pyproject.toml and pyodk/__version__.py with the new release version number.
Update the Central version in the README to reflect the version we test against.
Commit, push the branch, and initiate a pull request. Wait for tests to pass, then merge the PR.
Tag the release and it will automatically be published (see release.yml actions file).